Exclusive: OpenAI unveils blueprint for teen AI safety standards
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Lindsey Bailey/Axios
OpenAI on Thursday released a blueprint the company hopes lawmakers will use in crafting safety standards for teens using AI, per a copy first shared with Axios.
Why it matters: OpenAI is trying to get ahead of criticism — and shape the norms for teens' use — by touting its new safety mechanisms to policymakers.
- This comes as more states across the country consider AI safety laws and senators roll out a bill that would ban chatbots for minors.
Driving the news: OpenAI is under scrutiny for how the platform handles younger users' safety as high-profile litigation involving a teen who died by suicide after interacting with the chatbot continues.
- OpenAI is also trying to extend its user base by pushing into more schools, where these kinds of safety frameworks are required.
What they're saying: OpenAI maintains that teens should have access to "safe and trustworthy" AI, per the blueprint, but that they should also "be protected from its potential harms."
- "We believe ChatGPT should meet them where they are: the way ChatGPT responds to a 15-year-old should differ from the way it responds to an adult," the report reads.
The company lays out five suggestions for AI companies to follow on protections for teens, which OpenAI has announced it's developing:
- identify teens on the platform and treat them accordingly in age-appropriate manners;
- mitigate risks to minors through policies requiring that AI systems do not depict suicide or self-harm, that they prohibit intimate and violent content and that they do not encourage dangerous stunts or harmful body ideals;
- default to an under-18 experience if "there is doubt" about a user's age;
- give families parental controls over their kids' accounts; and
- embed features informed by the latest research on teens and AI.
The big picture: Age verification on tech platforms is notoriously tough, and often easy for kids to get around.
- Companies like OpenAI may also be subject to stricter requirements under Europe's Digital Service Act soon, so they benefit from integrating safety onto their platforms sooner.
